Screening for latent tuberculosis in Norwegian health care workers: high frequency of discordant tuberculin skin test positive and interferon-gamma release assay negative results

Tuberculosis (TB) presents globally a significant health problem and healthcare workers (HCW) are at increased risk of contracting latent TB infection (LTBI) and develop active TB [1, 2]. The prevalence of TB infection and disease are highest among HCW in low/middle income countries [1]. The overall incidence of TB in Norway is 7/100.000 and of the approximately 350 new TB cases annually 80-90% are imported infections from TB-endemic countries [3]. According to the national guidelines those exposed for TB, including HCW, should be screened for TB infection and preventive anti-TB chemotherapy should be considered [4].

The tuberculin skin test (TST) has high sensitivity, but low specificity, especially in Bacillus Calmette-Guérin (BCG)-vaccinated individuals, because of cross-reactivity to non-tubercular mycobacterias (NTM) or due to booster-effect caused by repetitive testing [5]. This has resulted in long-term medical follow-up for many HCW with positive TST. There is no diagnostic gold standard test for LTBI, but blood based interferon-gamma release assays (IGRA) offer better specificity (98–100%) and at least as good sensitivity (70–97%) as the TST since they are unaffected by previous BCG vaccination and most NTM [2, 6–9]. These assays, commercially available as QuantiFERON®TB Gold In-Tube (QFT) and T-SPOT.TB® measure in-vitro IFN-γ production by T-cells stimulated with the M. tuberculosis-specific antigens ESAT-6, CFP-10 and TB7.7. The tests have already been incorporated into national guidelines [10–13] and in Norway the QFT is recommended since 2007 as the test of choice following TST in a two-step approach [14]. Still, the role of IGRA in HCW screening is unclear [15].

Studies from TB low-endemic countries generally find a low prevalence of LTBI in HCW defined by IGRA [15, 16, 18–27]. A high proportion of discordant TST positive/IGRA negative results have been reported in HCW, especially in countries with high coverage of BCG vaccination [15, 16, 23, 27–29]. In Norway a limited number of studies of the performance of IGRA in various populations have been published [23, 30–33], but no study has focused on the routine screening of HCW. We have previously shown in a cohort with a predominance of immigrants that although one third tested QFT positive, in HCW, a subgroup of the study participants, only 10% had a positive QFT test [32].

We have performed a cross-sectional study amongst HCW in three major hospitals in the Western region of Norway to study the prevalence of QFT positivity in HCW working in hospital departments with possible risk of TB infection, identify possible risk factors for LTBI in our hospitals and investigate the performance of the QFT test in routine testing of HCW.

We present a cross sectional study on routine IGRA testing of HCW working in three major hospitals in the Western region of Norway, a TB low-endemic country. We find that only 3.4% of the HCW tested positive with QFT in contrast to 55.3% with TST, indicating low level of LTBI in HCW working at Norwegian hospitals. This is in support of a previous Norwegian contact investigation study that also found a frequency of 3% of LTBI among HCW by the T-SPOT.TB test [23] and a Danish study from hospital medical wards where only 1% of the HCW had a positive QFT [16]. In contrast, in a study from Russia, a TB high-endemic country, 47% of the HCW in the TB department were IGRA positive [17]. We did not find any effect of age on the QFT results. In the study by Nienhaus et al, however, IGRA results among HCW depended greatly on age and country with a QFT positivity of 25% in Germany, 45% in Portugal and 33% in France in the age group over 55 years [26]. In a recent review from Zwerling et al. the pooled prevalence of positive IGRA using either test was significantly lower than for a positive TST among HCW in low TB incidence settings, and IGRA showed good correlation with occupational risk factors for TB exposure [29].

The 13 HCW with a persistent positive QFT were equally distributed between the three hospitals and between nurses, medical doctors and laboratory staff and the majority was born in Norway. All HCW reported known exposure to TB at work, but had used infection control equipment during patient care and diagnostic procedures. Seven HCW reported additional possible environmental exposure to TB. Further, four of the HCW had either origin from and/or reported working in TB high-endemic regions. It is likely that these were infected abroad since country of origin was the only risk factor significantly associated with a positive QFT test in the multivariate analysis. One HCW had been diagnosed with active TB several years back illustrating that IGRAs could stay positive long after treated infection [32]. From 2005 a total of 47 HCW with TB infection, of whom 14 developed active TB, have been reported in Norway (personal communication, The Norwegian Institute of Public Health). However, 25 were born outside Norway possibly implying that TB exposure had occurred abroad. Based on our results we cannot conclude for sure, but altogether this indicates a low risk of contracting TB and a good infectious control at Norwegian hospitals.

The growing literature raises the question of whether IGRA could replace the TST when screening HCW for TB in routine practice or contact investigations due to the improved specificity [15, 18, 26–28]. This is especially relevant in countries like Norway with high coverage of BCG vaccination and discordant TST positive/IGRA negative results. In a Canadian study LTBI prevalence among HCW measured by the TST was low and the most common discordant test results were TST negative/QFT positive [35]. In contrast, we observed that a total of 33% of the HCW had TST > 10 mm, 13.7% had TST ≥ 15 mm with a frequency of only 4% QFT positivity and even HCW with the highest TST values > 25 mm were QFT negative. This might represent false negative QFT, but there was no association between known TB risk factors and TST ≥ 15 mm. BCG vaccination at the age of fourteen was performed according to the Norwegian national vaccination program from 1947 to 2009 and explains why 98% of all the HCW in our cohort were vaccinated. Thus, the discrepancy it most likely explained by false positive TST due to high BCG vaccination coverage or booster effects after repeatedly TST testing in the hospitals. With over half of the HCW in our study demonstrating a positive TST, which we believe is representative for HCW working at Norwegian hospitals, it is obvious the TST in our setting generally is of little use. Instead, IGRA offer better specificity and thereby lower prevalence of positive tests and fewer HCW who require X-rays, further clinical follow-up or LTBI treatment. Still, medical follow-up of HCW with strong TST reactions and risk factors for TB reactivation might be warranted since there is no diagnostic gold standard for LTBI.

There is concern about IGRA reproducibility, defining the optimum cut-off values for positive tests which most accurately distinguish new TB infection from random variation as well as defining the right interpretation of discordant TST and IGRA in serial screening [29, 36, 37]. In our study, the HCW with initial positive QFT that retested negative all had low initial IFN-γ responses below 0.70 IU/ml. Poor reproducibility of the assay is a more likely explanation than true reversions since the tests were performed within short periods. Such fluctuations in IGRA findings were also demonstrated in a study where HCW in contact with TB patients were tested monthly [38]. When an increase in QFT cut-off from < 0.35 to ≥ 0.70 IU/ml was applied inconsistent IGRA results were reduced from 52% to 27% and consistency in QFT results was associated with baseline IFN-γ levels. A recent review also concludes that subjects with baseline results around the diagnostic threshold are more likely to show inconsistent results on retesting [36]. Also in our cohort there was a tendency to higher IFN-γ levels in those with consistent positive QFT. Thus, a borderline zone from 0.20-0.70 IU/ml has been suggested in the routine screening of HCW with retesting before preventive chemotherapy is recommended [36]. Longitudinal research is therefore requested to inform guidelines on IGRA serial testing.

Some countries such as the USA have almost since their introduction recommended the two new IGRAs for screening of LTBI [11]. Others, including the Norwegian guidelines, have in general been more cautious and recommended a two-step approach with an initial TST [14]. The European guidelines conclude that IGRA may be used as part of the overall risk assessment to diagnose LTBI and identify individuals for preventive treatment [10]. In the updated guidelines from the USA the advantages and disadvantages of the various tests are discussed [39]. When choosing test one must consider the accuracy of the test in the specific risk groups as well as populations. The guidelines also conclude that due to the high negative predictive value, progression to active TB in healthy immune competent individuals with negative IGRA is very unlikely.

According to Norwegian guidelines HCW who has worked or stayed in a TB endemic country or are included in contact investigations should be screened for TB infection in a two-step approach with an initial TST before IGRA [14]. Many hospitals have also performed regular routine TST in HCW with risk for TB infection at their workplace. Our study demonstrating a high frequency of discordant TST positive/IGRA negative results indicates that the National guidelines should be updated and an “IGRA-only strategy” considered for HCW. Use of IGRA in this population is expected to increase diagnostic specificity and improve acceptance of treatment for LTBI. The higher per-test cost of IGRAs may be compensated for by lower post-screening costs (medical follow-up, chest x-rays and chemoprevention). A recent review of cost-effectiveness of IGRA concludes that in four studies, the “two-step strategy” and in two studies the “IGRA-only strategy” was most cost-effective [40]. A recent British study concludes that IGRA can be an institutional cost saving and result in higher compliance rates [41]. There is growing evidence in support of the use of IGRA in screening risk groups such as HCW. However, one must be aware of the uncertainty of IGRA during serial-testing with yet undefined proper cut-off levels and possible low-level false-positive IGRA results that will lead to unnecessary follow-up of low-risk HCW.

The limitations of our study are the small number of IGRA positive to draw strong conclusions on the overall prevalence of TB infection in HWC in Norway, the risk factors for TB infection and the effects of serial testing. In addition it is not possible to show the frequency of converters since no repetitive testing of the QFT negative were performed.

We find in this cross sectional study that only 3.4% of HCW were QFT positive in contrast to 55.3% TST positive indicating a low level of LTBI among HCW working at hospitals in Western Norway. The only risk factor associated with a positive QFT was origin from a TB-endemic country implying a low risk for contracting TB infection at Norwegian hospitals. The “IGRA-only” seems like a desirable strategy, despite its limitations, due to the high number of false positive TST probably caused by BCG vaccination and repetitive TST testing. Thus, there is a need for an update of Norwegian guidelines for TB screening in HCW in order to get more accurate diagnosis of LTBI and offer proper treatment and follow-up.